Memory systems and neural dynamics

Abstract

Philosophers, psychologists, and biologists have long pondered how experiences mark the mind. It is clear that there is no one seat of memory. Within each memory system there are multiple biological mechanisms for information storage. This chapter focuses on how memory could be supported by coincidence of neural activity, changes in synaptic strength, and synchronization of networks. Another major theme in memory research is the localization and specialization of function by different memory systems which can be revealed through appropriate behavioral tests. Recent technological advances allow for powerful new ways to record and perturb neural circuits to test the veracity of long-held theories about the relationship between biology and cognition. The fundamental challenge for a memory system is to recognize past experiences that have utility in guiding action. This is difficult because no two situations are identical, and therefore it is necessary for the brain to match which events and stimuli should be grouped together due to common meaning. This is the computation that occurs upon meeting someone new and thinking, “You remind me of an old friend." From this flash of familiarity, recollections of past experiences shape the interaction with the stranger and, in doing so, bias how this new person is seen and which aspects of the experience become committed to memory. This everyday example illustrates the profound influence memory has on learning, behavior, and even perception. There are several ways to seek an explanation for how the brain uses the past to bias future behavior. One is to design experiments that decrease the probability of an event’s occurrence, such as removing a part of the brain and observing memory loss. These loss-of-function experiments elucidate the necessity of a specific phenomenon for the occurrence of another. Other experiments are designed to cause something to happen, such as the delivery of a drug that enhances memory performance. These experiments inform us as to the sufficiency that the presence of one event is enough to observe another. Gain of function and loss of function experiments may perturb the system in unexpected ways, and therefore it is also important to observe the brain with minimal intervention. In such observations, experimenters seek to identify variables that covary, such as the activity of a neuron correlating with some stimulus. A final and important method for understanding the link between biology and cognition is computational and mathematical modeling of neural systems (Hopfield 1995; Marr 1971). These more abstract analyses can offer predictions that, if validated, lead credence to the model from which the prediction was derived, which may provide insight into processes beyond the current limit of experimental validation. Furthermore, there may be many biologic solutions to the same computational problem and modeling efforts help to identify necessary and sufficient overarching principles that unite seemingly disparate observations. The convergence of these lines of evidence - necessity, sufficiency, correlation, and modeling - has shown that there are multiple memory systems that specialize in storing and processing different kinds of information. In each of these systems, information is thought to be stored in changes in communication efficacy between neurons. The temporal organization of neural activity required for such changes in synaptic strength and for efficient communication is achieved by coordination of brain rhythms across brain regions. This chapter will handle these broad topics with an emphasis on the technological advances in behavioral testing, genetics, imaging, and recording that allowed each new scientific advance. Very few contemporary concepts in memory research are brand new, and therefore it is essential to first discuss the insights of the pioneers that have shaped the current understanding of the neurobiology of memory